# Neutron detection A [[Neutron|neutron]] is uncharged, so it leaves no ionisation track and cannot be detected by any method that works on a [[Proton|proton]], an [[Alpha_particle|alpha particle]] or an [[Electron|electron]]. Every neutron detector is therefore a two-stage machine — a conversion reaction that turns the neutron into charged particles, followed by an entirely conventional charged-particle detector — and the rest of the field, from the choice of fill gas to the shape of the pulse-height [[Histogram|spectrum]], the [[Polymer_engineering|polyethylene]] jacket on a portal monitor, and a supply crisis that raised the price of [[Helium-3|helium-3]] as much as fiftyfold, descends from that single fact. ## Microsims — three.js <iframe src="https://wikitube-3d-microsims.netlify.app/Neutron_detection.html" width="100%" height="620" frameborder="0" loading="lazy" sandbox="allow-scripts allow-same-origin" title="Neutron detection — three.js microsim"></iframe> **`Neutron_detection` (three.js).** The sim builds a ³He proportional counter — translucent cathode, anode wire on the axis — and nothing in it is faked: capture depth is sampled by inverse transform from exp(−nσx), so at high pressure the captures visibly pile against the entrance wall; the proton and triton launch back-to-back with their correct 572.6 and 191.3 keV; and the deposited energy of a clipped track comes from the residual-range relation, so the ion-pair markers trace dE/dx and reproduce the Bragg profile for free. Work the pressure and diameter sliders while a live pulse-height spectrum accumulates on the backboard from actual events — drop the pressure to 1 atm in a 1 cm tube and the 45 mm proton range exceeds the radius nine times over, so 100% of events are wall-clipped, the full-energy peak disappears entirely and only a continuum remains. Push the neutron energy to the top of its slider, about 1 MeV, and the 1/v law drives σ to 0.845 b: the neutrons pour straight through and the HUD reads TRANSPARENT TO THESE NEUTRONS. The live quantity to watch is the pair of efficiency readings — the predicted ε = 1 − exp(−nσL) printed alongside the measured captured/incident ratio — because you can sit and watch one converge on the other. ## The founding difficulty A [[Neutron|neutron]] feels no Coulomb [[Force|force]]. Undeflected by the [[Voltage|field]] of an atomic [[Electron|electron]] or a nucleus, it makes no [[Ion|ions]], no scintillation, no [[Electric_current|current]] — nothing any [[Sensor|sensor]], [[Transducer|transducer]] or [[Electronics|readout]] can register. Every technique in charged-particle [[Instrumentation_and_control_engineering|instrumentation]] reads out a trail of [[Ionization_energy|ionisation]]; a neutron leaves none, which puts it outside the ordinary apparatus of [[Detection_theory|detection theory]]. So a neutron detector is necessarily **converter plus detector**: a nucleus absorbs the neutron and promptly emits charged products, those ionise, the ionisation is amplified as usual. The instrument never sees the neutron, only the debris. Nuclide and [[Density|density]] fix efficiency; what the [[Radioactive_decay|reaction]] releases — a [[Binding_energy|binding-energy]] difference — fixes pulse height and hence [[Photon|gamma]] rejection; and where the converter *comes from* has mattered more than any of the [[Physics|physics]]. ## The conversion reactions Four converters carry the field. All are exothermic, so pulse height is set by the reaction's Q-value and not by the neutron's [[Energy|kinetic energy]]: a thermal neutron of 0.0253 eV yields a pulse of hundreds of keV. | Reaction | Q | Charged products | σ at 0.0253 eV | |---|---|---|---| | ³He(n,p)T | **763.8 keV** | p **572.6 keV** + T **191.3 keV** | **5333 ± 7 b** (Sears/NIST) / **5316 b** (standards) | | ¹⁰B(n,α)⁷Li | 2.790 MeV (gs, ~6%) · 2.312 MeV (⁷Li\*, ~94%, + 478 keV γ) | α 1.78 / 1.47 MeV + ⁷Li 1.01 / 0.84 MeV | 3835 ± 9 b / 3842.6 b | | ⁶Li(n,α)T | 4.783 MeV | α 2.06 MeV + T 2.73 MeV | 940 ± 4 b / 938.5 b | | ²³⁵U(n,f) | ~200 MeV, ~160 MeV as fragment kinetic energy | two heavy fission fragments | 584.3 b | The [[Helium-3|³He]] split is not fitted; it follows from [[Newton's_laws_of_motion|momentum conservation]]. A thermal neutron brings negligible momentum, so [[Proton|proton]] and triton leave back-to-back with equal and opposite momenta, and since E = p²/2m the Q-value divides in *inverse* proportion to the [[Atomic_mass|masses]]: > E_p = Q · m_T/(m_p + m_T) = 763.8 × 3.0155/4.0228 = **572.6 keV**, leaving the triton **191.3 keV**. Both round to 573 and 191 keV, which is fine. Quoting a cross-section with no neutron [[Velocity|energy]] attached is not, since the number moves over eight decades. The boron and lithium reactions divide their Q the same way, which is why their [[Alpha_particle|alpha particles]] carry the larger share. ## Why helium-3, and why it is trusted Two separable reasons. First, size: the thermal cross-section exceeds five thousand barns against a geometric nuclear area under one barn, so the [[Nucleosynthesis|nuclide]] is a target thousands of times larger than itself. Exothermic, no Coulomb barrier for an uncharged projectile, no resonances in the useful band — so σ obeys the **1/v law** almost perfectly: > σ(v) = σ₀·v₀/v, equivalently σ(E) = σ₀·√(E₀/E), with v₀ = 2200 m/s and E₀ = 0.0253 eV — the most probable speed of a room-temperature Maxwellian, out of the [[Kinetic_theory_of_gases|kinetic theory of gases]]. Second, [[Accuracy_and_precision|metrology]]. ³He(n,p)T is an *international neutron cross-section standard* from **0.0253 eV to 50 keV** — one of nine, alongside ⁶Li(n,t) and ¹⁰B(n,α) to 1 MeV and [[Carbon|C]](n,n) to 1.8 MeV. σ has been evaluated in a global least-squares fit against every competing measurement, so a ³He counter can *define* a flux rather than merely count in one. Hence its place in [[Nuclear_engineering|nuclear engineering]], [[Neutron_diffraction|neutron diffraction]] and safeguards: a [[Repeatability|reproducibility]] argument, not a sensitivity one. **The number to get right.** Two evaluations circulate and differ: **5333 ± 7 b** on the Sears/NIST table, **5316 b** in Carlson's cross-section *standards*. Both are legitimate and in wide use; cite the standards value for metrology, Sears for tabulated scattering data. The textbook "5330 b" the microsim uses sits between them. Wrong is the ubiquitous bare "5333 b", printed with neither its energy nor its evaluation. ## The proportional counter, and its first limit: efficiency The standard instrument is a gas proportional counter: a cylindrical [[Steel|steel]] cathode filled with ³He — the second [[Chemical_element|element]], and an [[Inert_gas|inert]] one — with the anode wire on the axis at one to two kilovolts. Electrons freed by the products drift and [[Diffusion|diffuse]] inward and, within a few wire radii where the field rises as 1/r, gain enough [[Energy|energy]] between collisions to multiply. One reaction makes 763.8 keV ÷ 42 eV ≈ **18,200 ion pairs** before gas gain, so the [[Signal|signal]] is large and the [[Signal-to-noise_ratio|signal-to-noise ratio]] excellent against amplifier [[Noise_(electronics)|noise]]. Efficiency is plain attenuation: a beam crossing gas of thickness L survives with [[Probability|probability]] exp(−nσL), so > **ε = 1 − exp(−nσL)** with n the number [[Density|density]] of ³He nuclei. Since n ∝ P for an ideal gas, efficiency is set by the *product* of pressure and path length, never either alone. At 0.0253 eV and 293 K, n = 2.47 × 10¹⁹ cm⁻³ per bar and Σ = nσ = 0.132 cm⁻¹ per bar — mean free path 7.6 cm at one bar. Makers quote **~90% of saturation at D·p ≈ 8 inch·bar**; direct calculation gives 93%. Typical fills are **2–10 bar**, and wall thickness, [[Alloy|alloy]] choice, weld [[Metallurgy|metallurgy]], [[Quality_assurance|quality]] control, [[Corrosion|corrosion]] resistance and [[Leak_detection|leak testing]] all descend from one [[Reliability_engineering|reliability]] requirement: hold several bar of an irreplaceable gas for decades. ## The second limit: the wall effect, and the floor pressure cannot lift If either product reaches the cathode wall before spending all its energy, the remainder is dumped into solid [[Materials_science|metal]] and never ionises. The event registers *below* the full-energy peak, and because the products fly back-to-back the spectrum grows two shelves: proton-escape events fill 191 keV to 763.8 keV, triton-escape events 573 keV to 763.8 keV. A clean line degrades into continuum — the dominant artefact of the instrument, and in the microsim two sliders destroy the peak outright. The instinct is to fix it with pressure: shorter tracks, fewer escapes. The instinct is wrong. > Σ = nσ ∝ P, while a charged particle's range at fixed energy R ∝ 1/P. Therefore **k = Σ·R is exactly pressure-independent.** k is the range measured in neutron mean free paths, and pressure crowds captures against the entrance wall by exactly as much as it shortens tracks. With ranges of ≈4.5 cm (proton) and ≈1.5 cm (triton) in pure ³He at 1 bar — helium being a [[Noble_gas|noble gas]] of low stopping power — **k_p ≈ 0.59 and k_T ≈ 0.20 at every pressure whatsoever**. Capture depth x follows the [[Probability_density_function|density]] Σe^(−Σx); an isotropically emitted product escapes back through that wall with probability ½(1 − x/R) for x < R; the [[Numerical_integration|integral]] closes in elementary form: > P_esc = ½·[ (1 − e^(−k)) − (1 − (1 + k)e^(−k))/k ] giving **12.3% for the proton and 4.6% for the triton** — disjoint populations, since the two fly oppositely — a near-wall escape floor of **roughly 15–17% that no amount of pressure can remove.** `[DERIVED]` A cylindrical [[Monte_Carlo_method|Monte Carlo]] of the microsim's geometry converges on 17.0–17.2% at large P·D and sharpens it: the clipped fraction depends *only* on P·D, the same group that sets efficiency. Neither a fatter tube nor a higher fill gets under the floor; only a heavy stopping gas — [[Krypton|krypton]], [[Carbon|C]]F₄ — shortens R without touching Σ and breaks the invariance. The ranges carry ~20% [[Estimation_theory|uncertainty]] and Bethe slowing-down ranges push the floor toward 22%, so read it as *of order one capture in six*, not a constant of nature. What the wall effect does not spoil is gamma rejection. A gamma in a few bar of low-atomic-number gas frees an [[Electron|electron]] that crosses the tube and leaves, depositing tens of keV at most — below even the bottom shelf of the neutron continuum. A discriminator above the gamma pile rejects [[Photon|photons]] by orders of magnitude while keeping nearly every neutron. That, more than raw efficiency, put ³He tubes into [[Security_engineering|security]] instruments standing in a [[Radium|radium]], [[Radon|radon]] and [[Thorium|thorium]] [[Decay_chain|decay-chain]] background. ## Fast neutrons, and what the polyethylene is for The 1/v law cuts both ways. Extrapolated to 1 MeV it gives **σ = 0.845 b** against 5333 b thermal — a factor near **6,300**, and the evaluated value there is about 0.9 b, so the law holds across the span. A bare ³He tube is close to transparent to fission-energy neutrons, as the microsim's HUD announces. Fast counting therefore means *thermalising first*. A [[Hydrogen|hydrogen]]-rich moderator, several centimetres of [[Polymer_engineering|polyethylene]], lets elastic scattering off protons drag the neutron down before it reaches the gas, since neutron and proton have nearly equal [[Atomic_mass|mass]] and exchange large [[Energy|energy]] fractions per collision. That is the plastic in every radiation portal monitor, and why such an instrument is bulky. It is also why these are counters, not spectrometers: the moderator deliberately destroys incident-energy information, so a [[Californium|²⁵²Cf]] [[Spontaneous_fission|spontaneous-fission]] source and an [[Americium|Am]]–[[Beryllium|Be]] source look alike downstream. A [[Cadmium|cadmium]] wrapper — ¹¹³Cd absorbs at 20,600 ± 400 b — is the complementary trick, cutting the thermal component to define an epithermal channel. ## The other detector families - **BF₃ proportional counters.** Boron trifluoride enriched to ~96% ¹⁰B is converter and counting gas at once, giving true gas multiplication and good gamma rejection. But [[Fluorine|fluorine]] is electronegative and attaches drifting electrons, so BF₃ runs at a fraction of an atmosphere and higher [[Voltage|voltage]] than ³He, and is toxic and corrosive. Lower σ times lower pressure is lower efficiency per unit length. - **[[Boron|Boron]]-lined proportional counters.** A micrometre-scale ¹⁰B or B₄C coating on the inner cathode, with ordinary [[Argon|argon]] counting gas. Only products emitted into the gas from within their escape depth count, so one layer converts at the percent level and the wall effect is intrinsic — continuum from the start. Efficiency is rebuilt by stacking layers or tubes: a [[Manufacturing|manufacturing]] problem, not a physics one. This family actually replaced ³He in portal monitors. - **⁶Li glass and ⁶Li scintillators.** [[Cerium|Cerium]]-activated lithium silicate glass (GS20 type) and ⁶LiF/ZnS(Ag) screens exploit the 4.783 MeV Q of ⁶Li(n,α)T — the largest non-fission Q available — for a bright pulse in a thin solid with fast timing. The cost is gamma sensitivity: a dense solid stops photons well, so discrimination needs pulse-shape [[Signal_processing|signal processing]] rather than a threshold. Commercial [[Lithium|lithium]] is often isotopically depleted, making enrichment a real expense. - **Fission chambers.** A thin [[Uranium|²³⁵U]] or [[Plutonium|²³⁹Pu]] deposit in an ionisation chamber. Fragments carry ~160 MeV, four orders of magnitude above anything a gamma deposits — the best gamma rejection of any neutron detector, which is why they instrument reactor cores where [[Nuclear_fuel|fuel]] monitoring and [[Safety_engineering|safety]] margins depend on it. Efficiency is low: the deposit must be thinner than the fragment range. - **Converter foils and spin filters.** [[Gadolinium|Gadolinium]] or ¹⁰B films on [[Silicon|silicon]] [[Semiconductor_device|semiconductor devices]] trade efficiency for spatial resolution in imaging. The same nuclide runs backwards too: polarised ³He, prepared by [[Hyperpolarization_(physics)|optical pumping]], absorbs in a strongly [[Spin_(physics)|spin]]-dependent way, so a gas cell becomes a neutron polariser — the [[Nuclear_magnetic_resonance|NMR]]-adjacent cousin of the counter above. Each trades against ³He on some axis — efficiency per unit volume, gamma discrimination, mechanical simplicity, toxicity, cost — and none dominates it. The reason the alternatives were built anyway is not [[Materials_science|materials science]]. It is [[Logistics|supply]]. ## The supply crisis: an instrument whose supply chain is a weapons artefact Essentially all commercial helium-3 on [[Earth|Earth]] is the [[Beta_decay|beta-decay]] [[Decay_product|product]] of tritium ([[Half-life|half-life]] **12.32 ± 0.02 y**, ~5.5%/y), harvested during maintenance of nuclear-weapons tritium reservoirs and separated by [[Fractional_distillation|cryogenic distillation]] — in the United States, NNSA Defense Program operations at the Savannah River Site, supplemented historically by Russian supply of roughly 25,000 L/y over 2004–2008. There is no significant terrestrial primary source: the [[Natural_gas|natural gas]] feeding the [[Helium_production_in_the_United_States|US helium industry]] and the [[National_Helium_Reserve|National Helium Reserve]] carries ³He at parts-per-billion of the [[Helium|helium]], hopelessly dilute against ordinary [[Helium-4|⁴He]], and no [[Helium_storage_and_conservation|conservation]] policy addressed the isotope separately. After **11 September 2001**, deployment of neutron-sensitive radiation portal monitors to catch smuggled special nuclear material — DHS fielded **over 1,400** — nearly tripled demand for a gas nobody was making on purpose, and whose production was if anything falling with the warhead stockpile after the [[Cold_War|Cold War]]. The shortage surfaced in **June 2008**, when the Spallation Neutron Source requested 35,000 L against an inventory that could not supply it; it became public in Congressional testimony in **November 2009**. The GAO audit put numbers on it. Extraction capacity was **8,000–10,000 L/y** against average sales of **~30,000 L/y** across 2003–2009 — three to four times production, drawn straight from stock. The **~260,000 L** available in 2003 fell to **~31,000 L** by February 2011, 209,888 L having been sold or transferred between. Prices went from **$40–85/L** at auction to **$365–1,000/L** allocated and up to **$2,000/L** commercially, in about two years. The finding worth quoting is not about geology: **no DOE entity had stewardship responsibility for helium-3.** The Isotope Program distributed it without controlling supply; NNSA extracted and held it without managing demand. Nobody owned the inventory curve — the government sold at three times its production rate for six years with no office whose job it was to notice. A failure of [[Systems_engineering|systems engineering]] and custody, not of [[Nuclear_engineering|nuclear engineering]]; the binding [[Limiting_factor|limiting factor]] was [[Availability|availability]], not sensitivity. The response was fast once it began, and it reshaped the landscape above: boron-lined and BF₃ alternatives were qualified for portal monitors, recycling and reclamation established, allocation moved to a White House interagency group, and federal demand is now projected **below 6,000 L/y** against the **70,000 L/y** peak of 2008. New supply appeared at the margin — Laurentis Energy Partners began extracting ³He from tritium stored at the Darlington CANDU station, announced 16 September 2021 as the first civilian, non-military source — while Russian supply has been effectively excluded from Western markets since 2022, and demand reopened as [[Dilution_refrigerator|dilution refrigerators]] for [[Quantum_computing|quantum computing]] and [[Superconducting_magnet|superconducting]] hardware multiplied, each holding a sealed charge of the [[Liquid_helium|liquid]] that gives [[Superfluidity|superfluid]] [[Helium_cryogenics|helium cryogenics]] its floor. Competing uses are small beside that: [[Hyperpolarization_(physics)|hyperpolarized]] ³He lung [[Magnetic_resonance_imaging|MRI]] has largely been displaced by ¹²⁹Xe on cost, and [[Aneutronic_fusion|aneutronic]] D–³He [[Nuclear_fusion|fusion]] remains prospective. Most speculative is the Bluefors–Interlune agreement of 16 September 2025: up to 10,000 L/y of *lunar* ³He from 2028 to 2037 — a purchase commitment against [[Regolith|regolith]] nobody has yet [[In_situ_resource_utilization|processed]], a [[Lunar_resources|resource]] rather than a reserve, on the [[Moon|Moon]]. Circulating 2025–26 market figures — production 22,000–30,000 L/y against demand 40,000–60,000 L/y, bulk price $1,900–2,600/L — rest on secondary compilations and are marked `[UNVERIFIED]` here. This is the best short case study anywhere of a scientific instrument's supply chain being an artefact of weapons policy. The standard detector of thermal neutrons — in reactor instrumentation, [[Neutron_diffraction|neutron scattering]] beamlines, safeguards, [[Petroleum_engineering|well logging]] and border security — runs on a gas that exists in quantity only because a state chose to build and maintain thermonuclear weapons, and became scarce because that state's response to a terrorist attack consumed it faster than its weapons programme made it. No market signal preceded the shortage, because there was no market: there was an inventory, and an unowned one. The [[Physics|physics]] of ³He(n,p)T has not changed since it was measured. What changed was custody. ## Sources - **Carlson, A. D. (2011).** "The neutron cross section standards, evaluations and applications." *Metrologia* **48**(6), S328–S345. DOI [10.1088/0026-1394/48/6/S09](https://doi.org/10.1088/0026-1394/48/6/S09). — Establishes ³He(n,p)T as an international standard over 0.0253 eV–50 keV (Table 3) and gives the standards thermal values used above: ³He 5316.00 b, ⁶Li(n,t) 938.47 b, ¹⁰B(n,α) 3842.56 b, ²³⁵U(n,f) 584.33 b (Table 4). - **NIST NCNR, neutron scattering lengths and cross sections** (Sears evaluation, *Neutron News* **3**(3), 1992, 29–37). [ncnr.nist.gov/resources/activation/scattering_table.html](https://www.ncnr.nist.gov/resources/activation/scattering_table.html) — The other widely used evaluation: ³He 5333(7) b, ⁶Li 940(4) b, ¹⁰B 3835(9) b, ¹¹³Cd 20600(400) b, σ_abs at 2200 m/s. - **CODATA 2022 fundamental constants**, NIST. [physics.nist.gov/cuu/Constants/Table/allascii.txt](https://physics.nist.gov/cuu/Constants/Table/allascii.txt) — The masses from which Q = 763.8 keV and the 572.6/191.3 keV split are derived, along with the ¹⁰B, ⁶Li and ²³⁵U Q-values tabulated above. `[DERIVED]` - **VacuTec, *Application: ³He Neutron Detectors*.** [PDF](https://www.vacutec-gmbh.de/fileadmin/VacuTec-Files/produkte/umwelt/Zaehlrohre__GM-__P-__N-_/Neutronen_Zaehlrohre/Application_He-3_Neutron_Detectors.pdf) — Confirms the Q-value as "764 keV", states the 1/√E dependence explicitly, and gives the practical fill range 2–10 bar and the ~90%-of-saturation-at-D·p = 8 inch·bar rule. - **U.S. Government Accountability Office (2011).** *Managing Critical Isotopes: Weaknesses in DOE's Management of Helium-3 Delayed the Federal Response to a Critical Supply Shortage.* **GAO-11-472**, 12 May 2011. [gao.gov/products/gao-11-472](https://www.gao.gov/products/gao-11-472) — Extraction capacity, sales and inventory figures, and the stewardship finding. - **Congressional Research Service (2010, rev. 2011).** *The Helium-3 Shortage: Supply, Demand, and Options for Congress.* **R41419**. [congress.gov PDF](https://www.congress.gov/crs_external_products/R/PDF/R41419/R41419.8.pdf) — Stockpile history (~140,000 L in 1990 → ~235,000 L peak in 2001 → ~50,000 L in 2010) and the price series. - **DOE National Isotope Development Center**, *Supply and Demand of Helium-3*. [isotopes.gov](https://www.isotopes.gov/Supply-and-Demand-of-Helium-3) — The official current position: mitigation by recycling and alternative technologies; federal demand now projected below 6,000 L/y. - **Wall-effect floor.** `[DERIVED]` — Analytic escape integral over the exponential capture-depth distribution at fixed k = Σ·R, cross-checked by a cylindrical Monte Carlo of the microsim geometry. Rests on ranges of ≈4.5 cm (proton) and ≈1.5 cm (triton) in pure ³He at 1 bar, themselves good to about 20%. - **2025–26 market figures** (world production 22,000–30,000 L/y; demand 40,000–60,000 L/y; $1,900–2,600/L). `[UNVERIFIED]` — secondary compilation citing Edelgas Group and the Lowy Institute; primaries not reached. ## Wikipedia : Wikitube **Strict pair:** [Wikipedia](https://en.wikipedia.org/wiki/Neutron_detection) : [Wikitube](https://en.wikitube.io/wiki/Neutron_detection) ## Previous hub tags Hubs: `Life_Physics`. Portals: [[PORTAL_Helium-3]].